【作者】 蔡远利； 石宏理；

【Author】 Cai Yuan-li Shi Hong-li (School of Electronic & Information Engineering,Xi’an Jiaotong University,Xi’an.Shanxi,710049)

【机构】 西安交通大学电子与信息工程学院；

【摘要】 离散小波变换为线性时变系统的辨识提供了一条新的途径,但现有算法运算量过大、无法实时在线实现．本文建立了一种基于小波分解的改进辨识算法,由于采用了递推计算格式,有效地避免了高维矩阵的求逆计算,可以构成一个实用的线性时变系统实时辨识框架。此外,我们还基于递代 EKF 的思想,改进了系统辨识中递推最小二乘参数估计算法,可以使辨识精度进一步提高．初步仿真表明,方法是有效、可行的。更多还原

【Abstract】 Discrete Wavelet Transform (DWT) has been applied to the identification of linear time-varying systems.However,the existing DWT approaches involve a great deal of computation burden so that the on-line applications are extremely difficult.In this paper,a new identification algorithm for linear time-varying systems is proposed based on the wavelet decomposition of the model parameters.The inversion computation of huge matrixes is productively avoided,therefore real-time implementation is now possible and a general identification frame for on-line applications is set up.Furthermore,a modified Recursive Least Square (RLS) algorithm embedded in the identification frame for parameter estimation is also given based upon the idea of iterated EKF,with which more accurate results are guaranteed.An illustrative simulation is presented to demonstrate the feasibility and efficiency of the proposed algorithms.更多还原